This is the whole course in one sitting. Every idea you’ve met — the cold paradox that costly helping should be bred out of existence, the shadow of the future that quietly rewrites the payoffs so cooperation becomes the selfish move, Axelrod’s four-line winner that never won a single duel yet won the war, the noise that traps two nice strategies in an endless feud, the five roads beyond simple reciprocity, and the honest catalogue of where the model lies — shows up here as questions. Hold the through-line in your head: cooperation among selfish agents is never magic, it is what selection builds the moment the game stops being one-shot and anonymous — when a future, a memory, a reputation, or a bloodline changes the payoffs. Take a breath. There is no going back once you commit — a fitting test for a course about the moves no one gets to unmake.
How this exam works
This is a real exam, not a practice quiz. Questions appear one at a time. Once you submit an answer it is locked for good — there is no going back, no retry, and no restart. Your score stays hidden until the very end. A few questions ask you to select all that apply (read those carefully — partial credit is not a thing here). You need 70% to pass. Ready when you are.
In a single, one-shot, anonymous prisoner's dilemma between strangers, what does cold self-interest prescribe, and why?
Select an answer to continue.
Big picture
The evolution of cooperation, in one picture
- The evolution of cooperation
- The paradox
- Blind selection rewards whatever out-breeds, so costly helping looks like suicide - a naive Always Cooperate world is invaded and erased by one Always Defect mutant, so it is not an ESS
- Shadow of the future
- Repeat the dilemma and a defection today costs all of tomorrow cooperation; when the continuation probability w clears the threshold, cooperating becomes the selfish move - but a known finite endpoint unravels it by backward induction
- How Tit-for-Tat won
- Axelrod ran a real tournament and the simplest four-line entry won: nice, retaliatory, forgiving and clear - it never won a single duel yet won the war by never losing badly, and nasty strategies poison their own well
- Noise and robustness
- One garbled move locks two Tit-for-Tat players into an endless revenge echo, and unforgiving Grim Trigger suffers most - refinements like generous and contrite Tit-for-Tat and Win-Stay-Lose-Shift (Pavlov) find the forgiveness sweet spot
- Beyond reciprocity
- Five roads beyond meeting again: indirect reciprocity and reputation (gossip), kin selection and Hamilton rB greater than C, network reciprocity (cooperators survive by clustering), and the contested multilevel group selection
- Conditions and limits
- Grow cooperation by lengthening the shadow, making actions visible, letting cooperators cluster, and changing the payoffs - not by moralising; and stay honest about fragility, the one-shot confusion, noise, and the naturalistic fallacy
- The paradox
Key takeaways
You now hold the whole model. The paradox is real: blind selection rewards whatever out-breeds, so costly helping () looks like suicide, and a naive Always Cooperate world is invaded and erased by one Always Defect mutant — it is not an ESS. The rescue is the shadow of the future: repeat the dilemma (, i.e. 5, 3, 1, 0) and a defection today forfeits all of tomorrow’s cooperation, so when the continuation probability w clears its threshold, cooperating becomes the selfish move — though a known, finite endpoint unravels it through backward induction (real, open-ended horizons escape). Axelrod ran a real tournament, and the simplest four-line entry, Tit-for-Tat, won by being nice, retaliatory, forgiving, and clear — it never won a single duel yet won the war by never losing badly, while nasty strategies poison their own well and go extinct once their prey runs out. Add noise and one garbled move traps two Tit-for-Tats in an endless feud (unforgiving Grim Trigger suffers most), which the refinements — generous and contrite Tit-for-Tat and Win-Stay-Lose-Shift (Pavlov) — are built to survive by hitting the forgiveness sweet spot. Beyond partners who meet again lie five roads: indirect reciprocity (reputation and gossip), kin selection (Hamilton’s rule rB > C, siblings , cousins ), network reciprocity (cooperators survive by clustering), and the contested multilevel group selection — never the naive “for the good of the species.” To grow cooperation you lengthen the shadow, make actions visible, let cooperators assort, and change the payoffs — not moralise. And stay honest about where the model lies: cooperation is fragile and invadable, the one-shot-versus-repeated confusion flips predictions, Tit-for-Tat is noise-fragile, and the naturalistic fallacy — mistaking what evolved for what is good — is a trap. Cooperation among selfish agents is not magic and not morality; it is what self-interest evolves into once the payoffs change.